Farr, Anne McCormack, authorWard, Robert C., advisorMcWhorter, David, committee memberSalas, Jose, committee memberBell, Harry, committee member2021-09-232021-09-231992https://hdl.handle.net/10217/233928This research focuses on the optimal design of groundwater quality monitoring networks. The optimization technique developed allows for incorporation of both the model structure, data error, and model parameter uncertainty into the monitoring network design, with concurrent determination of sampling frequency and well locations. Particular emphasis has been placed on the use of stochastic models to describe groundwater quality data in order to incorporate both the deterministic and random behavior of groundwater quality in the model evaluation and monitoring network design processes. A protocol is developed for the evaluation of model applicability and the design of monitoring networks. This protocol was developed based on the results of a simulation study, with the developed protocol tested against field data; The simulation study provided a method of evaluating the performance of various model applicability tests and monitoring network designs against a known correct model. The performance of the protocol could therefore be evaluated for correct models with different magnitudes and types of error (additive and multiplicative normal and lognormal errors were considered), as well as for incorrect models with different magnitudes and types of error. The results of this research strengthen the importance of a detailed statistical evaluation of model applicability prior to the use of a model as a tool for describing groundwater quality behavior or prior to the design of monitoring networks. The model applicability evaluation should include the use of a variety of statistical tests to assess model applicability, and more importantly, should include the evaluation of the behavior of statistical tests compared to the theoretical expected behavior of the statistical tests for a correct model under conditions of varying sampling frequency, record length, and sampling density. In addition, the optimal monitoring network was found to be highly dependent on the sampling locations used to fit the model and to the monitoring locations identified to be considered for inclusion in the monitoring network.doctoral dissertationsengCopyright and other restrictions may apply. User is responsible for compliance with all applicable laws. For information about copyright law, please see https://libguides.colostate.edu/copyright.Groundwater -- QualityWater quality managementWater quality monitoring stationsText